Saturday, September 13, 2014

Botany picture #174: Banksia serrata cone


A cone of Banksia serrata (Proteaceae), Australia, 2014. Often, seemingly simple and everyday things can lead to a number of intriguing questions when thought through. In this case, note first that there are very few follicles ("seed pods" - the red, hairy blobs in the picture) developing from what was a spike of dozens if not hundreds of flowers. In other words, most flowers were unsuccessful.

But why? Are these plants suffering from pollinator limitation, that is do they not get enough pollen to fertilise more flowers? Or do they get enough pollen, but most of it is from incompatible individuals? Many plants have incompatibility systems to ensure that they only accept pollen from sufficiently distantly related individuals, to avoid inbreeding. Maybe there just aren't enough compatible sex partners within a realistic distance, perhaps because the local population is genetically impoverished? Then again, maybe the plant could theoretically develop more genetically worthwhile fruits but it only ever "wants" to develop a few per cone because it cannot afford more, resource-wise, and so it aborts the rest. Offspring is expensive, and Proteaceae generally live on very nutrient-deficient soils.

Whatever it is, there is an interesting question here. I assume people will have studied it already, but I just don't know at the moment.

Another one is raised by the dry, dead flowers still attached to the cone. Some Banksias drop the dead flowers, e.g. B. integrifolia, some keep them. The latter state is often interpreted as an adaptation to frequent fire, but I recently heard an ecologist argue that many of the characters traditionally interpreted in that way arose before Australia dried out and became very fire-prone, so they couldn't have been adaptations. Well... don't characters always have to arise randomly before selection can begin to work on them? Where would you draw the line then?

That's the nice thing about the world around us: there is so much to figure out.

Wednesday, September 10, 2014

Multi-access keys, especially of the Lucid type

Long time no blog - I am somehow not finding the energy at the moment as often as I would like to. I would like to discuss a paper I recently found but before I can do that we need to lay some groundwork.

Previously I have discussed traditional dichotomous identification keys. They are the most familiar tool used for the identification of organisms, but there are less popular alternatives. Tabular keys are one option, and I have seen one a few years ago in a revision of the Bolivian species of a plant genus. It looked more or less like this:

Leaves <2 cm: P. vulgaris, P. aurantia, P. sericea
Leaves 2-5 cm: P. intermedia, P. reptans
Leaves >5 cm long: P. longifolia, P. boliviana, P. andina
Leaves glabrous: P. vulgaris, P. aurantia, P. andina
Leaves woolly: P. intermedia, P. reptans, P. longifolia
Leaves pilose: P. sericea, P. boliviana
...

Using a key like this is perhaps less intuitive and potentially a bit more confusing, but it can work quite well for relatively low species numbers. In the above case, for example, you would know immediately that you were dealing with P. andina if your sample had glabrous leaves longer than 5 cm because that species is the only one with that specific combination of characters. On the other hand, if you had very small glabrous leaves on your specimen it could still be either P. vulgaris or P. aurantia, so you should look for additional characters further down the list.

But as mentioned, this quickly becomes unrealistic for larger numbers of taxa; you wouldn't want to navigate such a table if every line listed more than twenty species possessing that particular character. Luckily, these days we have computers to help us with managing the data, and this is why tabular keys are becoming more popular.

The way this works in practice is that the taxonomist who wants to build a key produces a character-by-species table which the end user will not have to deal with. Instead, the end user is confronted with a software tool that allows them to enter whatever characters they can easily see, and each time they enter one the program throws out all the species that do not match the new information. The identification process ends when the possible species have been narrowed down to one or, more realistically a small number that the user may then check individually to see if one of them looks like the specimen they have on hand.

Perhaps the most popular software tool for building and displaying digital multi-access keys is Lucid, which, however, is not open source. Example identification keys for various groups can be found on their key server. The one that I am most familiar with is the Wattle Key to the Australian species of Acacia (Fabaceae). The screenshots below are accordingly taken from the Wattle Key, merely to demonstrate how Lucid Keys work:


There are four windows. The upper left one shows characters, the upper right one shows all the species, in this case 1274. The lower two windows are empty at the beginning. Now if we enter a few characters...


...we see that the characters that have been used appear in the lower left window. On the right, excluded species have moved to the lower right window while the ones that are still matching the available information are still in the upper right one. Our goal is to get their number down as far as possible.

The trick here is obviously to know which characters are more likely to be useful. In the present case, I pretended to have a bipinnate Acacia in front of me simply because I know that there are relatively few of them. Eh voila, as we see my two character selections already excluded all but 50 species.

This looks pretty straightforward, but of course if you don't know the plant group that well you won't know which characters might be important. In that regard a dichotomous key is more helpful because here the specialist who wrote it will have selected the most important characters for the first few questions of the key (if they are competent, that is). On the other hand, if you don't have the characters on your specimen that the first two questions of a dichotomous key ask for, for example because they are fruit characters but your specimen was still only in flower, then you are better served with a multi-access key.

When I next post about identification keys, some more of the pros and cons of both approaches will become obvious.

Friday, September 5, 2014

Botany picture #173: Cassytha glabella


A genus that I had never heard of before I came to Australia is Cassytha. As a twining, fully parasitic plant that has lost leaves and roots, it represents a remarkable case of convergent evolution with the dodder genus Cuscuta. Whereas the former is a member of the Lauraceae and as such of a very early diverging ("basal") lineage of flowering plants, the latter is deeply nested within the Asterid clade of the Eudicots. Shown here is fruiting Cassytha glabella, observed at Jervis Bay a few days ago. It appears to be parasitising on a member of the heath family Ericaceae.

Tuesday, September 2, 2014

Consciousness and the Fallacy of Composition

I wrote some time ago that I consider the Just World Fallacy to be perhaps the most pernicious fallacy there is, but lately it seems as if the Fallacy of Composition is popping up in discussions everywhere. Its influence is certainly not as destructive socially and politically, but it seems awfully widespread, and it seem to easily confound the thinking of otherwise smart and reasonable people.

What is the Fallacy of Composition? It is the mistake of concluding that the whole must have some property that its parts have individually or, perhaps more relevantly for present purposes, that the whole cannot have any properties beyond those of its parts. For example, one would be mistaken to conclude that birds cannot fly from the fact that individual feathers or leg bones are incapable of powered flight. So far, so obvious.

Where I increasingly notice the effect of this fallacy is in discussions of mind, consciousness and artificial intelligence. The funny thing about it is that the same mistake appears to be made by people on the most extreme ends of the spectrum, that is mind-body dualists and religious believers on one side and straight-laced rationalists and physicalists on the other side.

Monday, September 1, 2014

Waterfalls, this time with actual falling water

A few weeks ago I went to Jervis Bay with ANU lecturers and students for the third time. I always wanted to show the area to my family, and this weekend we finally managed to go camping there. My wife enjoyed the native plants and animals, my daughter loved the beach, and both admired the landscape.


On the way there, we stopped at Fitzroy Falls in the Moss Vale area. This impressive waterfall is 81 m deep, but as can be seen in the above picture there was quite a lot of fog so we could not see it under ideal conditions. This place is really wet most of the time, the vegetation is pretty much a temperate rainforest.


On the way back, we also stopped at the same place as I did with the student field trip - Tianjara Falls. Then I called it an alleged waterfall because it was dry, but this time we were rewarded with a much nicer view! For some time there was even a rainbow in the spray.


To end on a somewhat weirder note: The family posing in front of the Big Merino of Goulburn. These various big whatevers of Australia are of course greatly amusing to a five year old, but I wonder what archaeologists of the future are going to make of a megalithic sheep. If it can still be reconstructed after two thousand years it will probably be assigned some kind of religious significance. Then again, I doubt that a structure like this one will last as long as Roman masonry.

Thursday, August 28, 2014

Botany picture #172: Acacia decurrens


I am reasonably optimistic that this is indeed Acacia decurrens (Fabaceae), one of many wattles that are currently in flower here in Canberra. This was a small tree in Mount Majura Nature Reserve. It is the first time that I notice an Acacia with the styles of the individual flowers sticking out of the heads like that.

Wednesday, August 27, 2014

Final update on using fastStructure and similar software

After my somewhat mixed experience trying to use fastStructure, I have recently found the time to throw my data at two other programs for inferring population structure.

To recap, I have thousands of SNPs for two groups of species, in one case from 91 individuals and in the other from 224 individuals and I want to know how best to group the individuals into separate 'populations', in the present case potential species. I originally used fastStructure because it was new and supposedly written specifically for large numbers of SNPs, but the results were ultimately odd. The clusters didn't make very much sense and the program found virtually no admixed individuals, that is hybrids, although there really should have been some.

Earlier this week I then tried the R package adegenet. On the plus side, it turned out to be very simple and user-friendly. Of course you need to know how to use R, but the manual of the package is well written, and adegenet has a straightforward "read" function for importing datasets. It easily imported my Structure file without any hiccups, and after that it was a simple manner of handing my data over to adegenet's "find.clusters" function.

However, I tried different settings and did not get reasonable populations with any of them. One problem in my dataset are missing data, and I found that setting allele frequencies to zero for those cases produced the most meaningful results, but still there were several populations with no samples in them and the populations that had samples didn't make a lot of sense.

Yesterday I finally tried my luck with good old Structure itself - somewhat hesitatingly because I feared it would be very slow with such a big dataset. Yes, even for my smaller dataset what I wanted to do ran overnight, but that is still faster than I feared, and the results are worth it. The populations make sense, and in marked contrast to fastStructure it finds evidence of admixture. My larger dataset will probably need several days to be analysed, but if that is necessary so be it.

There is probably a reason why that program is the most popular in the area...